Accepted Manuscript

Factors affecting the status of food safety management systems in the global freshproduce chain

Klementina Kirezieva, Pieternel A. Luning, Liesbeth Jacxsens, Ana Allende, Gro S.Johannessen, Eduardo César Tondo, Andreja Rajkovic, Mieke Uyttendaele, MartinusA.J.S. van Boekel

PII: S0956-7135(14)00720-8

DOI: 10.1016/j.foodcont.2014.12.030

Reference: JFCO 4229

To appear in: Food Control

Received Date: 5 November 2014

Revised Date: 22 December 2014

Accepted Date: 23 December 2014

Please cite this article as: Kirezieva K., Luning P.A., Jacxsens L., Allende A., Johannessen G.S.,Tondo E.C., Rajkovic A., Uyttendaele M. & van Boekel M.A.J.S, Factors affecting the status of foodsafety management systems in the global fresh produce chain, Food Control (2015), doi: 10.1016/j.foodcont.2014.12.030.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

1

Factors affecting the status of food safety management systems in the global fresh 1

produce chain 2

Klementina Kirezievaa, Pieternel A. Luning a*, Liesbeth Jacxsensb, Ana Allendec, Gro S. 3

Johannessend, Eduardo César Tondoe, Andreja Rajkovicbf, Mieke Uyttendaeleb, Martinus 4

A.J.S van Boekela 5

6

a Food Quality and Design Group, Department of Agrotechnology and Food Sciences, Wageningen University, 7

P.O. Box 17, 6700AA Wageningen, The Netherlands 8

b Department of Food Safety and Food Quality, Laboratory of Food Preservation and Food Microbiology, 9

Faculty of Bioscience Engineering, University of Ghent, Coupure Links, 653, 9000 Ghent, Belgium 10

c Research Group on Quality, Safety and Bioactivity of Plant Foods, Department of Food Science and 11

Technology, CEBAS-CSIC, 30100 Murcia, Spain 12

d National Veterinary Institute, P.O.Box 750 Sentrum, 0033 Oslo, Norway 13

e Laboratório de Microbiologia e Controle de Alimentos, Instituto de Ciências e Tecnologia de Alimentos, 14

Universidade Federal do Rio Grande do Sul (ICTA/UFRGS), Av. Bento Gonçalves, 9500, prédio 43212, Campus 15

do Vale, Agronomia, Cep. 91501-970 Porto Alegre/RS, Brazil 16

f Department of Food Safety and Quality Management, University of Belgrade, Faculty of Agriculture, 17

Nemanjina 6, 11080 Belgrade-Zemun, Serbia 18

19

*Corresponding author. Mailing address: Food Quality and Design Group, Department of Agrotechnology 20

and Food Sciences, Wageningen University, P.O. Box 17, 6700AA Wageningen, The Netherlands, 21

Phone:+317482087. Fax: +317483669 Electronic mail address: Pieternel.Luning@wur.nl 22

23

Keywords: fresh produce; food safety management; Europe; emerging and developing 24

countries 25

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

2

Abstract 26

Increase in global trade raised questions regarding status of food safety management systems 27

in fresh produce companies, especially from developing and emerging countries. The aim of 28

this study was to investigate the status of food safety management systems (FSMSs) 29

implemented at primary production companies of fresh produce, to examine the potential 30

differences between companies operating in European Union (EU) and non-EU (developing 31

and emerging) countries, and to explore the underlying factors. Primary production 32

companies (n=118), located in the EU and in international cooperation partner countries 33

exporting to the EU, were assessed by using a diagnostic tool. The results from the study 34

indicated that several factors have a dominating effect on the status of FSMSs in the global 35

fresh produce chain. International export supply chains promote capacity building within 36

companies in the chain, to answer the stringent requirements of private brand standards. This 37

was shown to be an important factor in emerging and developing countries, where local 38

institutional environments often fail to support companies in setting and implementing their 39

FSMSs. Moreover, the legislative framework in these countries still requires improvements in 40

the establishment and enforcement. All this has negative consequences for the FSMSs in 41

companies supplying the local markets. In companies located in the EU, sector and other 42

produce organisations facilitate the sampling for pesticide residues and collaboration in the 43

sector. Overall, farmers showed less knowledge and overall awareness regarding 44

microbiological hazards, which is related to the less attention paid to these in the current 45

legislation and standards. Furthermore, standards are an important tool to trigger the 46

maturation of the systems as companies that were lacking any pressure to comply to standards 47

operated at a very basic level - with only few activities implemented. The insights from this 48

study indicate the need of stratified measures and policies to support companies in the fresh 49

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

3

produce chain in designing and operating their FSMSs according to the institutional 50

environment in which they operate. 51

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

4

1. Introduction 52

The world production of fresh produce raised by 38% in the last decade (FAOSTAT, 2013). 53

The demand for seasonal and exotic fruits and vegetables has also increased and trade with 54

fresh produce is more and more international (Diop & Jaffee, 2005). Simultaneously, food 55

safety problems linked to fresh produce have been repeatedly reported, for instance, recent 56

outbreaks with pathogenic microorganisms such as EHEC in sprouted seeds in Germany and 57

France, Listeria monocytogenes in melons in USA, and norovirus in berries in Nothern 58

Europe (EFSA, 2011; Laksanalamai, et al., 2012; Fründt, et al., 2013; Bernard, et al., 2014). 59

Moreover, breach of pesticide residue limits is a common problem (RASFF, 2012; Winter, 60

2012). Some of these food safety scares involved multiple sources and countries of 61

consumption, making it difficult to trace back the (single) point source of contamination in 62

the country of origin (Lynch, Tauxe, & Hedberg, 2009; EFSA, 2014). Fresh produce is often 63

imported from warmer climates located in developing and emerging economies, and these 64

were frequently associated with the problems (RASFF, 2012). 65

To address food safety, every company in the global food chain needs to implement a food 66

safety management system (FSMS) (CAC, 2009, 2010a). Each FSMS is company specific 67

because it is a result of the implementation of various quality assurance and legal 68

requirements into a company’s unique production, organisation and environment (Jacxsens, 69

Luning, et al., 2011). FSMSs implemented in companies at primary production are based on 70

good agricultural and good hygiene practices, by following national and international, public 71

and private standards and guidelines (CAC, 2010a; GlobalGAP, 2012). These are put into 72

place to assure product quality and safety across countries and regions. Certain countries and 73

regions however (e.g. the European Union) have put more stringent requirements about food 74

safety, which are difficult to reach by some companies in less developed countries (Unnevehr, 75

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

5

2000; Kussaga, Jacxsens, Tiisekwa, & Luning, 2014; Kussaga, Luning, Tiisekwa, & Jacxsens, 76

2014; Sawe, Onyango, & Njage, 2014). 77

Studies on the implementation status of FSMSs across industries of animal-derived products 78

in different countries and sectors highlighted that small and medium companies can also 79

demonstrate advanced systems (Djekic, et al., 2013; Luning, et al., 2015). A body of research 80

investigated the adoption of quality assurance standards in developing and emerging 81

economies as driven by export supply chains, and their role as barriers and facilitators to 82

economic growth (e.g. Jaffee & Henson, 2004; Henson, 2008; Henson & Humphrey, 2010; 83

Herzfeld, Drescher, & Grebitus, 2011). However, safety is still considered as less important 84

than price, quality and delivery conditions during the selection of suppliers (Voss, Closs, 85

Calantone, Helferich, & Speier, 2009). To address these issues, extra controls are established 86

in the EU for import products of non-animal origin with a history of safety problems, and 87

these include a number of products from emerging and developing countries (EC, 2009, 88

2010). Moreover, scientific evidence suggested that FSMSs implemented in fresh produce 89

companies in developing and emerging economies are not sufficiently addressing the food 90

safety risks (de Quadros Rodrigues, et al., 2014; Sawe, et al., 2014; Uyttendaele, Moneim, 91

Ceuppens, & El Tahan, 2014). However, deeper understanding is needed about the actual 92

status of FSMSs in different world countries involved in fresh produce trade. Moreover, 93

insight is lacking about the factors determining differences in the FSMS status, for instance 94

between companies in the EU and in (importing) developing and emerging countries. 95

The objective of this study was to investigate the status of FSMSs implemented at primary 96

production companies of fresh produce, to examine the potential differences between 97

companies operating in EU and non-EU (developing and emerging) countries exporting to the 98

EU, and to explore the explaining factors. 99

100

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

6

2. Materials and methods 101

2.1.Diagnostic tool for assessing the status of FSMSs in the fresh produce chain 102

The data for the study was collected with a recently developed diagnostic tool that allows 103

assessment of the status of FSMSs in the fresh produce chain, independently from type of 104

product and production, location, standards and guidelines used for the development of the 105

system (Kirezieva, Jacxsens, Uyttendaele, Van Boekel, & Luning, 2013; Kirezieva, 106

Nanyunja, et al., 2013). An FSMS is this part of the quality management system of a company 107

that is specifically addressing food safety (Luning, et al., 2009). The diagnostic tool allows 108

assessment of the core control and assurance activities in an FSMS (Kirezieva, Jacxsens, et 109

al., 2013), and the context factors (product, production, organisational and chain 110

characteristics) affecting design and operation of activities in the FSMS (Kirezieva, 111

Nanyunja, et al., 2013). Finally, the tool allows to measure the system output and the insight a 112

company has on its performance (e.g. results of external inspections or audits, results of 113

sampling) (Kirezieva, Jacxsens, et al., 2013). The diagnostic allows for assessment throughout 114

the FSMSs applied in the supply chain. In this work the diagnostic tool for farm level is 115

applied, including activities as part of the good agricultural practices or any other 116

implemented standards and guidelines. Indicators and grids with stereotypical situations are 117

defined for each indicator of the context, FSMS activities and FSMS output (Table 1). The 118

overall assumption behind the assessment is that high context riskiness requires an advanced 119

level of activities to achieve good output. 120

121

2.2.Data collection 122

Hundred and eighteen (118) companies from twelve (12) countries participated in the study 123

on voluntary ad-hoc basis. The data was collected within case studies, which were selected by 124

following the criteria of: vulnerability to food safety hazards (microbiological, pesticide 125

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

7

residues), economic relevance, vulnerability to climate change and consumption patterns and 126

trends (Jacxsens, Van Boxstael, & Uyttendaele, 2011). Companies involved were from 127

countries in the European Union (EU): Belgium (leafy greens and strawberries), the 128

Netherlands (leafy greens, strawberries and tomatoes), Spain (leafy greens); and the European 129

Economic Area: Norway (leafy greens and strawberries). The companies in this group of 130

countries were operating under the general food law of the European Union (EC, 2002). A 131

second group of companies were from international cooperation partner countries (ICPC): 132

Brazil (leafy greens), China (apples), Egypt (leafy greens and strawberries), India (mangoes), 133

Serbia (raspberries), Kenya (green beans), South Africa (leafy greens and fruits), and Uganda 134

(hot peppers). The companies in these countries operated under their own legal framework. 135

Data was collected with the diagnostic tool by interviewing the quality assurance manager or 136

farm owner for about one hour and a half by an on-farm visit by the researchers. In several 137

countries (the Netherlands, Spain, South Africa) workshops were organised to fill in the 138

diagnostic tool with help by the researchers. Part of the assessments were paired with visits 139

and microbiological sampling, which will be presented in other manuscripts. Definitive 140

conclusions about the actual food safety levels, however, should be made with caution, 141

because we could compare only with a limited number of companies. In India, China and 142

Egypt, data was collected via interviews with experts in the particular production sector. 143

These were people from national agencies, involved in the establishment of FSMSs at farm 144

level. The study cannot be conclusive for the countries involved, as the sample was not 145

representative for all sectors and supply chains. Contacts with companies were established via 146

the researchers in the countries involved, and based on voluntary participation. 147

148

2.3.Data analysis 149

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

8

A database was designed in Microsoft Excel 2010 with numbers given (1, 2, 3 and 4) to 150

represent companies’ situation for each of the 69 indicators. Descriptive statistics was 151

performed to determine frequencies and median scores for the companies operating under and 152

outside of the EU general food law. These were also compared statistically by using Mann-153

Whitney U nonparametric test, with significance of results established at p < 0.05. 154

Hierarchical cluster analysis of z-scores was performed by using the furthest neighbour 155

method. Further on, principal component analysis (PCA) with direct Oblimin method and 6 156

retained factors, was used to investigate the principal factors that explain the variation 157

between the companies and to explore which indicators are differentiating the three clusters. 158

Modes were calculated to plot the results in spider web diagrams. All statistical tests were 159

performed by using software package SPSS Statistics 21 for Windows. 160

161

3. Results 162

3.1.Overall status of FSMSs in companies under and outside the EU food law 163

Table 2A displays frequency distributions and median scores for indicators of FSMS 164

activities for the companies operating under and outside the EU food law. Significant 165

differences were reported in the design of several core preventive measures such as, storage 166

facilities, incoming materials control, water control, supplier control, pesticide program, 167

maintenance and calibration program, and irrigation method. The majority of EU companies 168

were following standards and guidelines (median score 3 for 9 out of 12 indicators), and using 169

best available tools and equipment, whereas the non-EU companies showed larger variation 170

and common basis for the activities was the use of own knowledge and experience (median 171

score 2 for 7 out of 12 indicators). Monitoring activities were mostly lacking in the non-EU 172

companies (all indicators with median score 1), but they were also not implemented in many 173

EU companies (Table 2A). This was especially the case with microbiological sampling 174

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

9

(frequency 32/69). Still, EU companies had better insights into actual operation of control 175

measures as indicators for availability and compliance to procedures and performance of 176

measuring equipment scored at average (3) and advanced level (4). Most companies, both in 177

and outside the EU, had limited information about actual hygienic performance of equipment 178

(1). Assurance activities in the majority of the EU companies were set up according to 179

standards and guidelines (median score 3 for 6 out of 9 indicators), while for the non-EU 180

companies they were missing (1) or at basic level (2). The exception was documentation 181

which scored 3 for companies in both EU and non-EU countries. 182

Table 2B presents the frequency distributions of the indicators scores of FSMS output for the 183

companies operating under and outside the EU general food law. All indicators scored 184

significantly different between companies in EU and non-EU countries, except for the 185

indicators of microbiological sampling and judgement criteria. Most companies did not have 186

information for these, as product sampling for microbiological analysis was not performed. 187

The majority of EU companies were audited by two and more third parties, while half of non-188

EU companies were not audited at all, and half - by only one-third party. Most EU companies 189

had a complaints management system and few customers’ complaints, whereas a smaller 190

percentage of the non-EU companies demonstrated the same results. Similarly, pesticide 191

sampling on a sector or company level was common in the EU, but not in the companies 192

outside of the EU. 193

Table 2C presents the frequency distributions and median of the results for context indicators 194

for the companies operating under and outside the general food law of the EU. Product 195

characteristics exhibited medium (2) to high-risk (3) scores, due to the nature of the 196

commodities included in this study, such as leafy vegetables and berries. Moreover, final 197

products are mostly eaten raw and possibilities for removal of both microbiological and 198

chemical contamination are limited. The features of the production system were also assessed 199

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

10

as high risk (3), as most companies were operating open field where contamination can occur 200

from people and environment (e.g. wild life). Climate conditions and water supply were 201

assessed mostly at moderate risk (score 2) for the companies in the EU, because they typically 202

operate in moderate climate zones and use underground water sources. The risk was high (3) 203

for the companies operating outside of Europe, as many were located in tropical and sub-204

tropical countries and were commonly using surface water. For all companies organisational 205

characteristics were mostly at moderate (2) and high risk (3), which represent technical staff 206

with limited knowledge on safety, use of seasonal workers with low competences and 207

involvement, and general quality/safety policy as e.g. introduced by the retailer. The majority 208

of EU companies operated in moderate riskiness (2) of context due to more formalization and 209

supporting information system. Bigger differences were observed between EU and non-EU 210

companies for the chain characteristics. Most EU companies showed lower risk of chain 211

context (5 indicators at score 2, and 3 at score 1) compared to the non-EU companies (5 212

indicators at score 4, and 3 indicators at score 2). This was because of the regular information 213

exchange, local supply, sophisticated logistic facilities, supportive food safety authority, 214

external support (such as sector organisations) and legal framework. However, they had to 215

answer stricter stakeholder requirements, compared to their non-EU counterparts. Common 216

for all companies was the moderate risk (2) of stakeholder requirements. 217

218

3.2.Clusters of companies differing in FSMS status 219

A hierarchical cluster analysis was performed to further explore how companies group 220

according to their FSMS status and adaptation of FSMSs to the riskiness of the context, which 221

resulted in three clusters (Figure 1). Table 3 is showing the distribution of the companies in 222

the clusters according to type of product, size, and certifications. Cluster 1 consists of forty-223

seven companies located in both EU and non-EU countries. These companies have 224

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

11

implemented and were certified against several voluntary private standards, such as 225

GlobalGAP, BRC, ISO, and private brand standards. In cluster 2 forty-two companies were 226

grouped from both EU and non-EU countries. However, these were certified only against 227

national standards and GlobalGAP, and no other private standards were included. In the last 228

cluster 3 twenty-nine companies were grouped, that were small and from non-EU countries. 229

These companies were not certified against any standard. 230

Figure 2 presents the median score results for the context, FSMS activities and output 231

indicators. Product and process characteristics scored similar for cluster 1 and 2 companies. 232

However, the lower score (1) for the indicator ‘microbial risk of initial materials’ was due to 233

treatments of seeds and seedlings (UV, chemical disinfection). Climate conditions and water 234

source scored 3 (high risk) for the companies in cluster 3, which was related to tropical and 235

sub-tropical conditions, and the use of uncontrolled surface water sources (see also Table 2C). 236

A similar pattern was observed for the indicators of organisational and chain characteristics, 237

as cluster 1 demonstrated least risky (scores 1 and 2), and cluster 3 - most risky profile 238

(mostly scores 3). The latter showed lack of organisational (i.e. lack of technical staff, low 239

competences of workers, lack of management commitment and workers’ involvement), and 240

supply chain support (i.e. lack of power in supply chain and information exchange, lack of 241

logistic facilities and support from authorities and other organisations). Still, low risk scores 242

(1) were given to the indicators of ‘variability of workers’, ‘severity of stakeholder 243

requirements’ and ‘specificity of food safety legal framework’. These scores can be explained 244

by the fact that cluster 3 contained small family farms, oriented mainly towards local markets 245

and ethnical shops in the EU countries. 246

FSMS activities of the companies in cluster 1 were mostly at level 3, design according to the 247

standards and using standard equipment and methods. Even more advanced level 4 was given 248

for storage facilities, fertilisation program, pesticide management, and sampling for pesticide 249

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

12

residues and microorganisms. Level 4 is attributed when activities are adapted to fit-for-250

purpose and tested for specific farm/company situation. Companies in cluster 2 scored mostly 251

at 3 (average level), but several activities such as hygienic design of equipment, material 252

control and irrigation method demonstrated even lower level (2). This score was given to 253

activities designed according to historical knowledge and own experience. Companies in 254

cluster 3 had basic control activities. Only few activities (these are, maintenance program, 255

personal hygiene requirements, and pesticide program) were actually implemented in these 256

companies. 257

Companies in cluster 1 had also average level (3) for the indicators about actual operation of 258

control activities, and even advanced level (4) for actual storage capacity and performance of 259

analytical equipment. Companies in cluster 2 did not have information about many of the 260

activities. Still, procedures were available, updated ad-hoc and mostly complied to. No 261

information about actual operation was available in the companies from cluster 3. 262

Assurance activities in companies of cluster 1 were at average level (3), meaning that 263

companies were actively following the changes, regularly updating the system, validating via 264

external experts, regularly verifying the activities internally, and systematically documenting. 265

Most of the activities were at basic level (2) in the companies in cluster 2, and validation was 266

not done (1). Still, systematic documentation and record-keeping were at place (3). Validation 267

activities were lacking completely for the companies in cluster 3. 268

Regarding the information for the system output, companies in cluster 1 were regularly 269

audited by several third parties for their multiple standards (3), had few remarks and 270

complaints from customers (4). These companies were taking systematic samples for 271

pesticide residues and even a sampling plan for microorganisms was present (3). Companies 272

in cluster 2 showed similar results for FSMS evaluation, complain registration and pesticide 273

sampling. However, they were lacking sampling plans microorganisms (1), and non-274

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

13

conformities registration (1). Again, companies in cluster 3 did not have any information 275

about their output (1). 276

277

3.3. Exploring the main factors behind differences between the clusters 278

A Principle Component Analysis (PCA) was performed to investigate the main factors 279

explaining the variation and differentiation between clusters of companies. Four indicators 280

were excluded because the Kaiser-Meyer-Olkin measure of sampling adequacy was below 281

0.5; namely ‘microbial risk of initial materials’, ‘risk of initial materials to pesticide residues’, 282

‘microbial risk of final product’, ‘risk of final product to pesticide residues’ and ‘variability of 283

workers’. Figure 3 presents the loading plots for the first three principle components. The first 284

component (PC1) explains 36.9%, the second component (PC2) - 7.3%, and the third 285

component (PC3) - 5.1% of variation (49.3% of the total variation). The next three 286

components explain below 5% of the variance, 4.1% (PC4), 3.6% (PC5) and 3.2 (PC6), and 287

they were not further investigated. 288

When riskiness of organisational and chain characteristics load negatively on PC1 (lower 289

risk), then design of several core control activities (these were, sanitation program, 290

maintenance program, supplier control, material control), actual operation of control activities 291

(availability of procedures and hygienic performance), and all assurance activities load 292

positively, with more advanced levels (Figure 3). This principle component may be linked to 293

availability of specific, scientific based information within the company and the supply chain 294

to support control and assurance activities. When requirements from stakeholders load as low 295

risk and support from authorities load as high risk on PC2, then adequacy of analytical 296

equipment for pesticide residues loads low (lower level), and packaging and partial 297

intervention load positively (higher levels). This principle component can be linked to the 298

support of public and private organisations in monitoring (of pesticide residues), providing 299

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

14

information about post-harvest intervention strategies and the factors affecting their 300

effectiveness. Storage loads positively on PC3, together with irrigation method, personal 301

hygiene and hazard analysis. This could be linked to companies that invest in storage and 302

related main control measures (e.g. export oriented). 303

PCA score plots were constructed with a total variance of 49.3% (PC1 and PC2; Figure 4A 304

and 4B) to demonstrate the differentiation between clusters. Cluster 1 was separated from 305

cluster 3 by PC1, while cluster 2 was somewhere in between and most companies were split 306

by PC2. However, eleven companies loaded positively on PC2 and ten of them were from 307

Norway. From Figure 4A it can be derived that cluster 1 companies were mostly separated 308

from the rest by PC1. Cluster 2 companies were fragmented by both PC1 and PC2. Figure 4B 309

shows that cluster 3 was split from cluster 1 by PC3, while companies in cluster 2 were again 310

scattered. 311

312

4. Discussion 313

• Status of FSMSs in view of context riskiness 314

The data was collected with the diagnostic tool, which allows assessment of the core control 315

and assurance activities in an FSMS, the system output (Kirezieva, Jacxsens, et al., 2013), and 316

the context factors affecting design and operation of activities in the FSMS (Kirezieva, 317

Nanyunja, et al., 2013). The general assumption behind the diagnostic tool used for data 318

collection is that companies working in high-risk context need advanced level of activities to 319

achieve a good system output (Luning, et al., 2011). The combination of high risk-context 320

characteristics and simple (basic level) FSMS imply a higher risk of food safety problems (de 321

Quadros Rodrigues, et al., 2014; Uyttendaele, et al., 2014). 322

All companies in this study were dealing with risky product linked to type of cultivated 323

produce, open field production and potential microbiological or chemical contamination of 324

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

15

fresh produce, and high risk organisational characteristics, such as high turnover of workers 325

with low involvement and competences. These are typical issues that fresh produce farms face 326

(Ahumada & Villalobos, 2011). However, most EU companies were working in a lower or 327

moderate risk of production and supply chain context, compared to the non-EU companies 328

(Table 2C). This was due to the use of controlled water sources, protected cultivation, 329

sophisticated infrastructure, and collaborative supply chain. Protected cultivation and 330

controlled water sources are used to increase yields, control abiotic factors and promote pest 331

management in integrated fresh-cut and other added-valued supply chains (Nicola, et al., 332

2009). 333

Many non-EU companies had simple FSMS based on own knowledge and experience, which 334

in combination with their high-risk context characteristics induce a higher risk on safety 335

problems (Table 2A). Previous studies discuss shortcomings in FSMSs leading to food safety 336

problems (Kussaga, Jacxsens, et al., 2014; Uyttendaele, et al., 2014). In comparison, most 337

companies within the EU had control activities based on standards, using expert knowledge 338

and standard equipment (Table 2A). Studies done in developing and emerging countries were 339

discussing the lack of knowledge and competences in the companies, particularly 340

smallholders (García Martinez & Poole, 2004; Trienekens & Zuurbier, 2008; Kussaga, 341

Jacxsens, et al., 2014; Sawe, et al., 2014). Assurance activities were following similar trend 342

with the exception of validation of monitoring systems and verification of people, which were 343

mostly ad-hoc in the EU companies and lacking in many non-EU companies. A study in the 344

meat and dairy sectors in Europe showed similar difficulties in setting verification and other 345

assurance activities (Sampers, Toyofuku, Luning, Uyttendaele, & Jacxsens, 2012; Jacxsens, et 346

al., 2015; Luning, et al., 2015). It was stressed before that verification is crucial for assuring 347

effectiveness of activities, but requires knowledge and resources (Nguyen, Wilcock, & Aung, 348

2004; Luning, et al., 2009). 349

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

16

Following the basic assumption behind the diagnostic tool, the EU companies showed better 350

system output, and generally more information about it was available to them (Table 2B). 351

They were audited by two and more third parties, which was not done or only by one in the 352

non-EU companies. This could be explained by the fact that audits in developing countries are 353

commonly done by the importers (retailers) only in companies aimed at exporting (Jaffee & 354

Masakure, 2005; Henson & Jaffee, 2008). Multinational retail chains have a dominating role 355

in imposing quality and safety standards and in many situations the role of the local 356

institutions is still weak (Berdegué, Balsevich, Flores, & Reardon, 2005). 357

In general, EU companies had also more information about the output of their FSMSs due to 358

company sampling and sector monitoring for pesticide residues. In developing countries that 359

was mainly done in the export farms, as local governments and organisations lack 360

infrastructure, capacity and resources to conduct monitoring (Jaffee & Henson, 2005; 361

Abhilash & Singh, 2009). Pesticide residues have been regulated in the EU (EC, 2005), and 362

by Codex Alimentarius Commission (CAC, 2010b) for some years now. Companies have put 363

lots of efforts in the management of pesticides, and this was evident also from our study as 93 364

(out of 118) companies demonstrated average (3) and advanced levels (4) (Table 2A). In 365

comparison, microbiological hazards only recently received some attention in (international) 366

recommendations, as a reaction to outbreaks e.g. sprouted seeds - EU 2011 (Soon, Seaman, & 367

Baines, 2013), spinach - USA 2006 (CFERT, 2007), and Jalapeño peppers - USA 2008 368

(Barton Behravesh, et al., 2011). Moreover, guidelines and standards for core control 369

activities to prevent microbiological contamination such as control of irrigation water quality 370

exist only in several regions and there is no internationally accepted recommendation for the 371

quality of water (Pachepsky, et al., 2011). Still most companies do not have enough 372

knowledge and awareness about microbiological hazards, and few of the companies in this 373

study performed sampling (score 1). 374

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

17

However, not all non-EU companies were facing the hurdles discussed above. A group 375

(clusters 1 and 2; Table 3) of non-EU companies also demonstrated average to advanced 376

FSMSs (Figure 2). Actually, non-EU companies were present in all the three clusters of 377

companies with differing FSMS status. After the cluster analysis typical profiles of companies 378

irrespective of their location and legislation followed (EU or not) were distinguished, and 379

several main factors that define these profiles were defined. 380

• The role of supply chains and private standards 381

From the study it became clear that collaborative/supportive supply chains (PC1; figure 3 and 382

4) contribute to more advanced FSMS and good system output (cluster 1; figure 2), as 383

companies demonstrated advanced knowledge and expertise about safety and quality 384

management. The results were independent from company size (micro, small, medium or 385

big), and location (EU or non-EU). However, the companies in this cluster were part of 386

leading fresh produce supply chains (in Spain, the Netherlands, Kenya, and South Africa). 387

Moreover, these companies had several certifications (Table 3), including many strict private 388

standards. Studies reported that some exporters have invested in infrastructure, quality 389

management and even own product-testing laboratories in order to meet strict requirements of 390

the (EU) markets (Jaffee & Masakure, 2005; Okello & Swinton, 2007; Henson, 2008). In 391

contrast, the companies in cluster 3 did not have any standards implemented and were 392

operating in less demanding chains supplying the local market or ethnic grocery shops in the 393

EU. They had only few control activities implemented scoring at basic level, which means 394

that they were following own knowledge and experience. No assurance activities and no 395

information about the system output were available. These results were linked to the lack of 396

information and expertise within companies and supply chains. Similarly, Trienekens & 397

Zuurbier (2008) report difficulties in the implementation of GAP and GHP in least developed 398

countries, and strive for implementation of international standards in emerging export 399

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

18

countries. Moreover, our findings are in line with studies demonstrating the lower levels of 400

safety and quality of horticultural products sold at wet markets and local supermarkets in 401

developing countries (e.g. Ponniah, et al., 2010; Gorton, Sauer, & Supatpongkul, 2011). 402

All companies in cluster 2 were following national standards or GlobalGAP, which is in line 403

with empirical evidence showing that prescriptive national standards lead to average levels of 404

FSMS activities, but companies still experience difficulties to tailor to their specific 405

circumstances (Aggelogiannopoulos, Drosinos, & Athanasopoulos, 2007; Sampers, et al., 406

2012). Interestingly this was the case also with the companies that were only GlobalGAP 407

certified, indicating the baseline status of this standard. Baseline standards put minimum 408

requirements, focusing on core activities from a public health perspective (Fearne & Garcia 409

Martinez, 2005). 410

411

• The role of the institutional environment 412

Sector organisations and non-governmental organisations (NGOs) that support companies 413

were another important factor affecting the status of FSMSs, especially in the case of small 414

and medium companies. These were largely represented in clusters 2 and 3. In both clusters 415

activities, such as, hygienic design of equipment, initial material control, and monitoring, 416

were not implemented. Even basic activities were lacking in cluster 3, and there was no or 417

only ad-hoc information about actual operation of control activities in both clusters. 418

Companies in cluster 2 and 3 had limited insights about the output of their systems (Table 2B) 419

due to e.g. shortcomings in monitoring of pesticide residues. In some EU countries these 420

activities are done or coordinated by sector organisations, but improvements are possible in 421

the structure and scientific base of the, e.g., sampling activities (Kirezieva, et al., 2015). 422

Coordinated sampling for pesticide residues was not done in many of the companies located 423

in emerging countries in clusters 2 and 3 (Figure 2). They were well established in the 424

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

19

companies from cluster 1, which were part of leading export chains and certified against 425

international standards. Henson, Masakure, and Cranfield (2011) defined two key factors for 426

companies in developing countries to acquire GlobalGAP certification: 1) being in an 427

established export country and 2) receiving technical and financial assistance. If companies 428

are not part of export supply chains, and support by foreign importers is lacking, then the 429

room for support is left to the (local) sector organisations or NGOs. However, in many cases 430

these are missing, dysfunctional or covering only staple crops production (Berdegué, et al., 431

2005; Nanyunja, et al., 2015). 432

Furthermore, the promulgation of a legal framework and its enforcement are the very 433

foundation for a good functioning FSMS, and especially important in the absence of 434

supporting (export) supply chain. This was demonstrated by the risky context due to low 435

requirements of stakeholders and restricted operation of local food safety authority (PC2), 436

which was linked to basic FSMSs (cluster 1). Companies in all the three clusters testified that 437

a legal framework that follows the international recommendations of Codex Alimentarius is in 438

place in their country (Figure 2). This is indeed so, as many countries have implemented the 439

requirements of Codex Alimentarius into their legal framework, but still having difficulties 440

with risk assessment, and with establishing sound policies and enforcement strategies that 441

consider the local social, technical and economic circumstances (Ecobichon, 2001; Chen, 442

2004; Abhilash & Singh, 2009). Moreover, evidence from emerging and developing countries 443

suggests that weaknesses in the institutional environment hinder the enforcement (Trienekens 444

& Zuurbier, 2008; Kussaga, Jacxsens, et al., 2014; Kussaga, Luning, et al., 2014). 445

446

5. Conclusions 447

This study demonstrated that several factors have a dominating effect on the status of primary 448

production FSMSs in the global fresh produce chain. International export supply chains 449

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

20

promote capacity building within companies in the chain, to answer the stringent 450

requirements of private brand standards. This is especially an important factor in emerging 451

and developing countries where local institutional environments often fail to support 452

companies in setting and implementing their FSMSs. Moreover, the legislative framework in 453

these countries still requires improvements in the set-up and enforcement. All this has 454

negative consequences for the FSMSs in companies supplying the local markets. 455

In companies located in the EU, sector and other produce organisations facilitate the sampling 456

for pesticide residues and collaboration in the sector. However, farmers showed less 457

knowledge and overall awareness regarding microbiological hazards, which is related to the 458

less attention paid to these in the current legislation and standards. Furthermore, standards are 459

an important tool to trigger the maturation of the systems as companies that were lacking any 460

operated at a very basic level - with only few activities implemented. 461

The insights from this study indicate the need of stratified measures and policies to support 462

companies in the fresh produce chain in designing and operating their FSMSs according to the 463

institutional environment in which they operate. This study was a first attempt to provide 464

evidence about major factors affecting the status of FSMSs in the fresh produce chain based 465

on semi-quantitative data analysis. More in-depth case studies into supporting contexts could 466

provide understanding about best strategies and practices to improve the status of FSMSs. 467

468

Acknowledgements 469

This research has received funding from the European Community’s Seventh Framework 470

Program (FP7) under grant agreement no 244994 (project Veg-i-Trade ‘Impact of Climate 471

Change and Globalization on Safety of Fresh Produce – Governing a Supply Chain of 472

Uncompromised Food Sovereignty’ www.veg-i-trade.org). We would like to thank the 473

experts and the companies that participated in the study. 474

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

21

References 475

Abhilash, P. C., & Singh, N. (2009). Pesticide use and application: An Indian scenario. Journal of 476 Hazardous Materials, 165(1–3), 1-12. 477

Aggelogiannopoulos, D., Drosinos, E. H., & Athanasopoulos, P. (2007). Implementation of a quality 478 management system (QMS) according to the ISO 9000 family in a Greek small-sized winery: 479 A case study. Food Control, 18(9), 1077-1085. 480

Ahumada, O., & Villalobos, J. R. (2011). A tactical model for planning the production and distribution 481 of fresh produce. Annals of Operations Research, 190(1), 339-358. 482

Barton Behravesh, C., Mody, R. K., Jungk, J., Gaul, L., Redd, J. T., Chen, S., Cosgrove, S., Hedican, 483 E., Sweat, D., Chávez-Hauser, L., Snow, S. L., Hanson, H., Nguyen, T.-A., Sodha, S. V., 484 Boore, A. L., Russo, E., Mikoleit, M., Theobald, L., Gerner-Smidt, P., Hoekstra, R. M., 485 Angulo, F. J., Swerdlow, D. L., Tauxe, R. V., Griffin, P. M., & Williams, I. T. (2011). 2008 486 Outbreak of Salmonella Saintpaul Infections Associated with Raw Produce. New England 487 Journal of Medicine, 364(10), 918-927. 488

Berdegué, J. A., Balsevich, F., Flores, L., & Reardon, T. (2005). Central American supermarkets’ 489 private standards of quality and safety in procurement of fresh fruits and vegetables. Food 490 Policy, 30(3), 254-269. 491

Bernard, H., Faber, M., Wilking, H., Haller, S., Höhle, M., Schielke, A., Ducomble, T., Siffczyk, C., 492 Merbecks, S. S., Fricke, G., Hamouda, O., Stark, K., Werber, D., & Team, o. b. o. t. O. I. 493 (2014). Large multistate outbreak of norovirus gastroenteritis associated with frozen 494 strawberries, Germany, 2012 Euro surveillance : bulletin europeen sur les maladies 495 transmissibles = European communicable disease bulletin, 19(8), 1-9. 496

CAC. (2009). Food hygiene. Basic texts. . In (Fourth Edition ed.). Rome, Italy: World Health 497 Organization, Food and Agriculture Organization of the United Nations. 498

CAC. (2010a). Code of hygienic practice for fresh fruits and vegetables. CAC/RCP 53-2003. In: 499 Codex Alimentarius Commission. 500

CAC. (2010b). Codex Pesticides Residues in Food Online Database. In: FAO/WHO Food Standards 501 Codex Alimentarius. Accessed on: 18 September 2014. 502

CFERT. (2007). Investigation of an Escherichia coli O157:H7 outbreak associated with Dole pre-503 packaged spinach. In: California Food Emergency Response Team. 504

Chen, J. (2004). Challenges to developing countries after joining WTO: risk assessment of chemicals 505 in food. Toxicology, 198(1–3), 3-7. 506

de Quadros Rodrigues, R., Loiko, M. R., Minéia Daniel de Paula, C., Hessel, C. T., Jacxsens, L., 507 Uyttendaele, M., Bender, R. J., & Tondo, E. C. (2014). Microbiological contamination linked 508 to implementation of good agricultural practices in the production of organic lettuce in 509 Southern Brazil. Food Control, 42(0), 152-164. 510

Diop, N., & Jaffee, S. (2005). Chapter 13. Fruits and vegetables: Global trade and competition in fresh 511 and processed produc markets. In M. A. Aksoy & J. C. Beghin (Eds.), Global Agricultural 512 Trade and Developing Countries (pp. 237-257): The International Bank for Reconstruction 513 and Development / The World Bank. 514

Djekic, I., Tomic, N., Smigic, N., Tomasevic, I., Radovanovic, R., & Rajkovic, A. (2013). Quality 515 management effects in certified Serbian companies producing food of animal origin. Total 516 Quality Management & Business Excellence, 25(3-4), 383-396. 517

EC. (2002). Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 518 January 2002 laying down the general principles and requirements of food law, establishing 519 the European Food Safety Authority and laying down procedures in matters of food safety. 520 Official Journal of the European Union, L 31, 1–24 521

EC. (2005). Regulation (EC) No 396/2005 of the European Parliament and of the Council of 23 522 February 2005 on maximum residue levels of pesticides in or on food and feed of plant and 523 animal origin and amending Council Directive 91/414/EEC (Text with EEA relevance) 524 Official Journal of the European Union, L 070 0001 - 0016. 525

EC. (2009). Commission Regulation No.669/2009 of 24 July 2009 implementing Regulation (EC) No 526 882/2004 of the European Parliament and of the Council as regards the increased level of 527

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

22

official controls on imports of certain feed and food of non-animal origin and amending 528 Decision 2006/504/EC. In E. Commission (Ed.). 529

EC. (2010). Commission Regulation No.1099/2010 of 26 November 2010 amending Annex I to 530 Regulation (EC) No 669/2009 implementing Regulation (EC) No 882/2004 of the European 531 Parliament and of the Council as regards the increased level of official controls on imports of 532 certain feed and food of non-animal origin. In E. Commission (Ed.). 533

Ecobichon, D. J. (2001). Pesticide use in developing countries. Toxicology, 160(1–3), 27-33. 534 EFSA. (2011). Tracing seeds, in particular fenugreek (Trigonella foenum-graecum) seeds, in relation 535

to the Shiga toxin-producing E. coli (STEC) O104:H4 2011 Outbreaks in Germany and 536 France. In Report of: European Food Safety Authority. 537

EFSA. (2014). Tracing of food items in connection to the multinational hepatitis A virus outbreak in 538 Europe EFSA Journal, 12(9), 186pp. 539

FAOSTAT. (2013). Food and Agriculture Organization Corporate Statistical Database. In 540 http://faostat3.fao.org/home/index.html#HOME. 541

Fearne, A., & Garcia Martinez, M. (2005). Opportunities for the coregulation of food safety: Insights 542 from the United Kingdom. Choices: The Magazine of Food, Farm, and Resource Issues, 543 20(2), 109-116. 544

Fründt, T. W., Höpker, W. W., Hagel, C., Sperhake, J. P., Isenberg, A. H., Lüth, S., Lohse, A. W., 545 Sauter, G., Glatzel, M., & Püschel, K. (2013). EHEC-O104:H4-Ausbruch im Sommer 2011. 546 23(5), 374-382. 547

García Martinez, M., & Poole, N. (2004). The development of private fresh produce safety standards: 548 implications for developing Mediterranean exporting countries. Food Policy, 29(3), 229-255. 549

GlobalGAP. (2012). Integrated Farm Assurance (IFA) Standard Version 4 550 Gorton, M., Sauer, J., & Supatpongkul, P. (2011). Wet Markets, Supermarkets and the “Big Middle” 551

for Food Retailing in Developing Countries: Evidence from Thailand. World Development, 552 39(9), 1624-1637. 553

Henson, S. (2008). The role of public and private standards in regulating international food markets. 554 Journal of international Agricultural Trade and Development, 4(1), 63-81. 555

Henson, S., & Humphrey, J. (2010). Understanding the Complexities of Private Standards in Global 556 Agri-Food Chains as They Impact Developing Countries. The Journal of Development 557 Studies, 46(9), 1628-1646. 558

Henson, S., & Jaffee, S. (2008). Understanding Developing Country Strategic Responses to the 559 Enhancement of Food Safety Standards. World Economy, 31(4), 548-568. 560

Henson, S., Masakure, O., & Cranfield, J. (2011). Do Fresh Produce Exporters in Sub-Saharan Africa 561 Benefit from GlobalGAP Certification? World Development, 39(3), 375-386. 562

Herzfeld, T., Drescher, L. S., & Grebitus, C. (2011). Cross-national adoption of private food quality 563 standards. Food Policy, 36(3), 401-411. 564

Jacxsens, L., Kirezieva, K., Luning, P. A., Ingelrham, J., Diricks, H., & Uyttendaele, M. (2015). 565 Measuring microbial food safety output and comparing self-checking systems of food 566 business operators in Belgium. Food Control, 49(0), 59-69. 567

Jacxsens, L., Luning, P. A., Marcelis, W. J., van Boekel, T., Rovira, J., Oses, S., Kousta, M., Drosinos, 568 E., Jasson, V., & Uyttendaele, M. (2011). Tools for the performance assessment and 569 improvement of food safety management systems. Trends in Food Science & Technology, 22, 570 Supplement 1(0), S80-S89. 571

Jacxsens, L., Uyttendaele, M., Devlieghere, F., Rovira, J., Gomez, S. O., & Luning, P. A. (2010). Food 572 safety performance indicators to benchmark food safety output of food safety management 573 systems. International Journal of Food Microbiology, 141, Supplement(0), S180-S187. 574

Jacxsens, L., Van Boxstael, S., & Uyttendaele, M. (2011). Selection of case studies based on the 575 vulnerability of fresh produce to food safety hazards and climate change in a globalized world. 576 In European Symposium of International Association of Food Protection. Ede, the 577 Netherlands. 578

Jaffee, S., & Henson, S. (2004). Standards and Agro-Food Exports from Developing Countries: 579 Rebalancing the Debate. World Bank Policy Research Working Paper, 3348. 580

Jaffee, S., & Henson, S. (2005). Chapter 6. Agro-food exports from developing countries: The 581 challenges posed by standards. In M. A. Aksoy & J. C. Beghin (Eds.), Global Agricultural 582

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

23

Trade and Developing Countries (pp. 91-114): The International Bank for Reconstruction and 583 Development / The World Bank. 584

Jaffee, S., & Masakure, O. (2005). Strategic use of private standards to enhance international 585 competitiveness: Vegetable exports from Kenya and elsewhere. Food Policy, 30(3), 316-333. 586

Kirezieva, K., Jacxsens, L., Hagelaar, G., van Boekel, M. A. J. S., Uyettendaele, M., & Luning, P. A. 587 (2015). Exploring the influence of context on food safety management: Case studies of leafy 588 greens production in Europe. Food Policy, under review. 589

Kirezieva, K., Jacxsens, L., Uyttendaele, M., Van Boekel, M. A. J. S., & Luning, P. A. (2013). 590 Assessment of Food Safety Management Systems in the global fresh produce chain. Food 591 Research International, 52(1), 230-242. 592

Kirezieva, K., Nanyunja, J., Jacxsens, L., van der Vorst, J. G. A. J., Uyttendaele, M., & Luning, P. A. 593 (2013). Context factors affecting design and operation of food safety management systems in 594 the fresh produce chain. Trends in Food Science & Technology, 32(2), 108-127. 595

Kussaga, J. B., Jacxsens, L., Tiisekwa, B. P. M., & Luning, P. A. (2014). Food safety management 596 systems performance in African food processing companies: a review of deficiencies and 597 possible improvement strategies. Journal of the Science of Food and Agriculture, n/a-n/a. 598

Kussaga, J. B., Luning, P. A., Tiisekwa, B. P. M., & Jacxsens, L. (2014). Challenges in Performance 599 of Food Safety Management Systems: A Case of Fish Processing Companies in Tanzania. 600 Journal of Food Protection, 77(4), 621-630. 601

Laksanalamai, P., Joseph, L. A., Silk, B. J., Burall, L. S., L. Tarr, C., Gerner-Smidt, P., & Datta, A. R. 602 (2012). Genomic Characterization of Listeria monocytogenes Strains Involved in a Multistate 603 Listeriosis Outbreak Associated with Cantaloupe in US. PLoS ONE, 7(7), e42448. 604

Luning, P. A., Bango, L., Kussaga, J., Rovira, J., & Marcelis, W. J. (2008). Comprehensive analysis 605 and differentiated assessment of food safety control systems: a diagnostic instrument. Trends 606 in Food Science & Technology, 19(10), 522-534. 607

Luning, P. A., Kirezieva, K., Hagelaar, G., Rovira, J., Uyttendaele, M., & Jacxsens, L. (2015). 608 Performance assessment of food safety management systems in animal-based food companies 609 in view of their context characteristics: A European study. Food Control, 49, 11-22. 610

Luning, P. A., Marcelis, W. J., Rovira, J., van Boekel, M. A. J. S., Uyttendaele, M., & Jacxsens, L. 611 (2011). A tool to diagnose context riskiness in view of food safety activities and 612 microbiological safety output. Trends in Food Science &Technology, 22, Supplement 1(0), 613 S67-S79. 614

Luning, P. A., Marcelis, W. J., Rovira, J., Van der Spiegel, M., Uyttendaele, M., & Jacxsens, L. 615 (2009). Systematic assessment of core assurance activities in a company specific food safety 616 management system. Trends in Food Science & Technology, 20(6-7), 300-312. 617

Lynch, M. F., Tauxe, R. V., & Hedberg, C. W. (2009). The growing burden of foodborne outbreaks 618 due to contaminated fresh produce: risks and opportunities. Epidemiology and Infection, 619 137(Special Issue 03), 307-315. 620

Nanyunja, J., Jacxsens, L., Kirezieva, K., Kaaya, A. N., Luning, P. A., Rademaker, M., & Uyttendaele, 621 M. (2015). Benchmarking food safety management systems for fresh produce in East Africa: 622 A case study of green bean farms in Kenya versus hot pepper farms in Uganda. Journal of 623 Food Protection, under review. 624

Nguyen, T., Wilcock, A., & Aung, M. (2004). Food safety and quality systems in Canada: An 625 exploratory study. International Journal of Quality & Reliability Management, 21(6), 655-626 671. 627

Nicola, S., Tibaldi, G., Fontana, E., Wojciech, J. F., Robert, L. S., Bernhard, B., & Prussia, S. E. 628 (2009). Chapter 10 - Fresh-cut Produce Quality: Implications for a Systems Approach. In 629 Postharvest Handling (Second Edition) (pp. 247-282). San Diego: Academic Press. 630

Okello, J. J., & Swinton, S. M. (2007). Compliance with International Food Safety Standards in 631 Kenya's Green Bean Industry: Comparison of a Small- and a Large-scale Farm Producing for 632 Export. Applied Economic Perspectives and Policy, 29(2), 269-285. 633

Pachepsky, Y., Shelton, D. R., McLain, J. E. T., Patel, J., Mandrell, R. E., & Donald, L. S. (2011). 634 Chapter two - Irrigation Waters as a Source of Pathogenic Microorganisms in Produce: A 635 Review. In Advances in Agronomy (Vol. Volume 113, pp. 73-138): Academic Press. 636

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

24

Ponniah, J., Robin, T., Paie, M. S., Radu, S., Ghazali, F. M., Kqueen, C. Y., Nishibuchi, M., 637 Nakaguchi, Y., & Malakar, P. K. (2010). Listeria monocytogenes in raw salad vegetables sold 638 at retail level in Malaysia. Food Control, 21(5), 774-778. 639

RASFF. (2012). The Rapid Alert System for Food and Feed: Annual Report 2012. In. Luxembourg: 640 RASFF. 641

Sampers, I., Toyofuku, H., Luning, P. A., Uyttendaele, M., & Jacxsens, L. (2012). Semi-quantitative 642 study to evaluate the performance of a HACCP-based food safety management system in 643 Japanese milk processing plants. Food Control, 23(1), 227-233. 644

Sawe, C. T., Onyango, C. M., & Njage, P. M. K. (2014). Current food safety management systems in 645 fresh produce exporting industry are associated with lower performance due to context 646 riskiness: Case study. Food Control, 40(0), 335-343. 647

Soon, J. M., Seaman, P., & Baines, R. N. (2013). Escherichia coli O104: H4 outbreak from sprouted 648 seeds. International Journal of Hygiene and Environmental Health, 216(3), 346-354. 649

Trienekens, J., & Zuurbier, P. (2008). Quality and safety standards in the food industry, developments 650 and challenges. International Journal of Production Economics, 113(1), 107-122. 651

Unnevehr, L. J. (2000). Food safety issues and fresh food product exports from LDCs. Agricultural 652 Economics, 23(3), 231-240. 653

Uyttendaele, M., Moneim, A. A., Ceuppens, S., & El Tahan, F. (2014). Microbiological safety of 654 strawberries and lettuce for domestic consumption in Egypt. Journal of Food Processing & 655 Technology, 5-3. 656

Voss, M. D., Closs, D. J., Calantone, R. J., Helferich, O. K., & Speier, C. (2009). The role of security 657 in the food supplier selection decision. Journal of Business Logistics, 30(1), 127-155. 658

Winter, C. K. (2012). Pesticide Residues in Imported, Organic, and “Suspect” Fruits and Vegetables. 659 Journal of Agricultural and Food Chemistry, 60(18), 4425-4429. 660

661

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table 1: Mutual characteristics of the different situations and levels in the diagnostic tool

Context factora Low risk (score 1)b Moderate risk (score 2) High risk (score 3) Product and production characteristics

Low chance of microbial or chemical contamination, growth or survival of pathogens and undesired microorganisms

Potential chance of microbial or chemical contamination, growth or survival of pathogens and undesired microorganisms

High chance of microbial or chemical contamination, growth or survival of pathogens and undesired microorganisms

Organisational characteristics

Supportive administrative conditions for appropriate decision-making.

Constrained (restricted) administrative conditions for appropriate decision-making

Lack of administrative conditions for appropriate decision-making

Chain characteristics Low vulnerability or dependability on other chain actors

Restricted vulnerability or dependability on other chain actors

High vulnerability or dependability on other chain actors

Food safety managementc Basic level (score 2) Moderate level (score 3) Advanced level (score 4) Control activities Standard equipment, unknown capability,

use of own experience/general knowledge, incomplete methods, restricted information, lack of critical analysis, and non-procedure-driven activities, regular unexpected problems, unstable

Based on expert (supplier) knowledge, use of (sector, governmental) guidelines, best practices, standardised, sometimes problems, causes known

Use of specific information, scientific knowledge, critical analysis, procedural methods, adapted and tested in the specific production circumstances, stable, robust

Assurance activities Problem driven, only checking, scarcely reported, not independent positions

Active translation of requirements, additional analysis, regular reporting, and expert support

Pro-active translation of requirements, actual observations and testing, independent positions, scientifically underpinned, and systematic

System outputd Poor output (score 2) Moderate output (score 3) Good output (score 4) Ad-hoc sampling, minimal criteria used for

evaluation, and having various food safety problems

Regular sampling, several criteria used for evaluation, and having restricted food safety problems mainly due to one (restricted) type of problem

Systematic evaluation, using specific tailored criteria, and having no safety problems

a Based on Luning et al. (2011) and (Kirezieva, Nanyunja, et al., 2013) b For all control and assurance activities and the system output, a level 1 is included to address situation when the activity is not applied (low level), or no information is available about the system output. c Based on Luning et al. (2008; 2009) and (Kirezieva, Jacxsens, et al., 2013) d Based on Jacxsens et al. (2010) and (Kirezieva, Jacxsens, et al., 2013)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table 2A: Frequency of the individual scores and median for the FSMS activities for EU and non-EU companies (The scores represent 1 – low level, 2 – basic level, 3 - average level, and 4 – advanced level), most frequent score indicated in bold.

FSMS activities EU (n=69) Non-EU (n=49) EU Non-EU

1a 2 3 4 1 2 3 4 median Preventive measures design Sophistication of hygienic design of equipment & facilities 19 15 29 6 41 7 1 3 1 Specificity of maintenance program* 4 9 43 13 8 24 7 10 3 2 Adequacy of storage facilities* 9 4 36 20 29 12 4 4 3 1 Specificity of sanitation program 12 18 28 11 18 13 14 4 3 2 Extent of personal hygiene requirements* 17 37 15 9 26 12 2 3 2 Sophistication of incoming materials control* 6 34 21 8 24 12 11 2 2 2 Adequacy of packaging 47 4 10 8 34 5 7 3 1 1 Sophistication of supplier control* 6 20 39 4 15 15 15 4 3 1 Specificity of fertilizer program 35 4 15 15 15 19 13 2 1 2 Specificity of pesticide program* 3 4 25 37 5 13 24 7 3 2 Sophistication of water control* 5 8 33 23 24 10 13 2 3 3 Adequacy of irrigation method* 24 39 6 15 21 12 1 3 2 Intervention method design Adequacy of partial physical intervention (e.g. disinfection) 40 9 15 5 25 16 6 2 1 1 Monitoring system design Appropriateness of hazard analysis* 14 11 38 6 29 8 10 2 3 1 Adequacy analytical methods for microbiological hazards 41 1 2 25 39 4 6 1 1 Adequacy of analytical methods for pesticide residues* 15 2 52 30 5 14 4 1 Specificity of microbiological sampling plan* 32 9 14 14 35 3 10 1 2 1 Specificity of pesticides' sampling plan* 21 6 30 12 37 3 9 3 1 Adequacy of measuring equipment* 12 3 28 26 41 1 7 3 1 Extent of corrective actions* 10 19 24 16 32 1 13 3 3 1 Actual operation of control activities Availability of procedures* 5 9 30 25 20 9 14 6 3 2 Compliance to procedures* 4 8 41 16 13 18 15 3 3 2 Actual hygienic performance of equipment & facilities 33 3 19 14 32 4 8 5 2 1

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Actual cooling and storage capacity* 19 9 18 23 36 1 5 7 3 1 Actual capability of partial intervention 51 2 10 6 42 1 6 1 1 Actual capability of packaging 51 4 7 7 37 1 5 6 1 1 Actual measuring equipment performance* 12 3 28 26 37 1 7 4 3 1 Actual analytical equipment performance* 25 1 1 42 39 2 2 6 4 1 Assurance activities Sophistication translating external requirements* 2 18 32 17 17 14 12 6 3 2 Extent of systematic use of feedback information* 7 20 28 14 21 17 6 5 3 2 Sophistication validating preventive measures* 15 18 32 4 24 12 11 2 3 2 Sophistication validating intervention strategies 39 10 17 3 31 7 9 2 1 1 Sophistication validating monitoring system 26 14 27 2 28 11 9 1 2 1 Extent verifying people related performance* 5 34 16 14 31 8 9 1 2 1 Extent verifying equipment & methods performance* 4 30 25 10 25 10 11 3 3 1 Appropriateness documentation* 2 11 43 13 19 5 21 4 3 2 Appropriateness record-keeping system* 1 12 48 8 16 11 19 3 3 2

a Situations 1, 2, 3, and 4 for control and assurance activities correspond to: low level (1) → absent, not applicable, unknown; basic level (2) → lack of scientific evidence, use of company experience/history, variable, unknown, unpredictable, based on common materials/equipment; average level (3) → best practice knowledge/equipment, sometimes variable, not always predictable, based on generic information/guidelines for the product sector; advanced level (4) → scientifically underpinned (accurate, complete), stable, predictable, and tailored for the specific food production situation. * Significant differences (p<0.05) between scores of EU and non-EU companies

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table 2B: Frequency of the individual scores and median for the FSMS output for EU and non-EU companies (The scores represent 1 – no information, 2 – poor output, 3 - moderate output, and 4 – good output), most frequent score indicated in bold.

FSMS activities EU (n=69) Non-EU (n=49) EU Non-EU 1a 2 3 4 1 2 3 4 median

Comprehensiveness external evaluation* 3 9 28 29 25 21 3 3 1 Seriousness of remarks* 3 1 6 59 27 4 8 9 4 1 Type of microbiological complaints* 21 3 8 37 23 8 8 10 4 2 Type of chemical food safety complaints* 8 1 21 39 21 7 13 8 4 2 Type of visual quality complaints* 6 13 41 9 16 14 13 6 3 2 Advancedness of microbiological sampling* 39 5 9 16 36 8 5 1 1 Comprehensiveness of judgement criteria for microbial FS 38 5 15 11 34 8 6 1 1 1 Advancedness of pesticides sampling* 14 5 39 14 34 4 7 4 3 1 Comprehensiveness of judgement criteria for chemical FS* 14 10 19 26 31 7 7 4 3 1 Type of non-conformities* 20 3 28 18 29 3 13 4 3 1

a Situations 1, 2, 3, and 4 for system output correspond to: no information (1) → absent, not applied, unknown; poor output (2) → ad-hoc sampling, minimal criteria used for evaluation, various food safety problems due to different problems in the activities; moderate output (3) → regular sampling, several criteria used for evaluation, restricted food safety problems mainly due to one (restricted) type of problem in the activities; good output (4) → systematic evaluation, using specific criteria, no safety problems. * Significant differences (p<0.05) between scores of EU and non-EU companies

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table 2C: Frequency of the individual scores and median for the contextual factors for EU and non-EU companies (The scores represent 1 – low risk, 2 – moderate risk, and 3 - high risk), most frequent score indicated in bold.

Context factors EU (n=69) Non-EU (n=49) EU Non-EU

1a 2 3 1 2 3 median Product characteristics Microbiological risk of initial materials* 11 19 39 10 26 13 3 2 Risk of initial materials to pesticide residues 18 60 1 6 33 10 2 2 Risk of initial materials due to mycotoxins* 1 33 35 20 22 7 3 2 Microbiological risk of final product* 1 7 61 1 15 33 3 3 Risk of final product to pesticide residues 4 12 53 5 2 42 3 3 Production characteristics Susceptibility of production system* 17 17 35 1 48 3 3 Risk climate conditions of production environment* 9 51 9 11 38 2 3 Susceptibility of water supply* 10 45 14 8 12 29 2 3 Organisational characteristics Presence of technological staff 15 30 24 5 17 27 2 3 Variability of workforce composition* 22 38 9 14 19 16 2 2 Sufficiency competences of operators 5 32 32 2 19 28 2 3 Extent of management commitment* 14 44 11 8 16 25 2 3 Degree of employee involvement 12 34 23 5 17 27 2 3 Level of formalization* 16 45 8 7 13 29 2 3 Sufficiency of supporting information system* 10 49 10 10 10 29 2 3 Chain characteristics Severity of stakeholder requirements* 10 45 14 21 24 4 2 2 Extent of power in supplier relationships* 10 44 15 5 19 25 2 3 Degree of information exchange in supply chain* 21 38 10 10 11 28 2 3 Sophistication of logistic facilities* 33 27 9 4 20 25 2 3 Supportiveness of food safety authority* 31 26 12 12 10 27 2 3 Degree of globalization of supply* 44 20 5 13 21 15 1 2 Specificity of external support* 60 7 2 13 7 29 1 3 Specificity of food safety legal framework* 69 24 11 14 1 2

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

a Low (1), medium (2) and high (3) risk situations for product and process characteristics correspond to low, potential and high chance of microbiological or chemical contamination. Ffor organizational and chain characteristics they correspond to supportive, constrained and lacking administrative chain conditions or low, restricted and high dependence on other chain actors. * Significant differences (p<0.05) between scores of EU and non-EU companies

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table 3: Distribution of the companies in the clusters

Country Product Sizea Standard

Leafy greens

Berries Fruits Other Micro Small Medium Large National Global GAP

BRC ISO Other

Cluster 1 (n=47) EU – Belgium (1) 1 1 1 1 – Netherlands (17) 3 6 1 7 4 4 9 15 2 4 – Spain (16) 13 3 1 4 12 3 16 5 7 7 Non-EU – Kenya (9) 9 4 5 5 9 1 5 – South Africa (4) 3 1 1 3 4 5 3

Cluster 2 (n=42) EU – Belgium (14) 6 8 8 5 1 14 14 – Netherlands (8) 4 3 1 4 4 8 – Norway (10) 6 4 1 5 4 10 1 – Spain (1) 1 1 1 1 Non-EU – Serbia (3) 3 2 1 3 – South Africa (6) 1 5 1 2 3 6 Cluster 3 (n=29) Non-EU – Brazil (6) 6 6 – China (2) 2 2 – Egypt (1) 1 1 – India (1) 1 1 – Kenya (1) – Serbia (6) 6 1 5 – Uganda (10) 10 10

TOTAL 41 37 10 27 20 49 27 20 34 78 7 10 16

a Size of companies is defined according to Commission recommendation of 6 May 2003 concerning the definition of micro, small and medium-sized enterprises (2003/361/EC).

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Cluster 3

Cluster 2

Cluster 1

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPTFigure 1: Dendrogram analysis of individual scores of 118 companies and identification of three clusters

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Product and process characteristicsa Organisational and chain characteristicsb Control activities designc

Clu

ster

1 (

n=46

)

Clu

ster

2 (

n=42

)

Clu

ster

3 (

n=29

)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Control activities operation Assurance activities System outputd

Clu

ster

1 (

n=46

)

Clu

ster

2 (

n=42

)

Clu

ster

3 (

n=29

)

Figure 2: Median score results for the indicators in the diagnostic instrument for the three identified clusters of farms active in fresh produce production

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

a Low (1), medium (2) and high (3) risk situations for product and process characteristics correspond to low, potential and high chance of microbiological or chemical contamination. b Low (1), medium (2) and high (3) risk situations for organizational and chain characteristics they correspond to supportive, constrained and lacking administrative chain conditions or low, restricted and high dependence on other chain actors. c Situations 1, 2, 3, and 4 for control and assurance activities correspond to: low level (1) → absent, not applicable, unknown; basic level (2) → lack of scientific

evidence, use of company experience/history, variable, unknown, unpredictable, based on common materials/equipment; average level (3) → best practice knowledge/equipment, sometimes variable, not always predictable, based on generic information/guidelines for the product sector; advanced level (4) → scientifically underpinned (accurate, complete), stable, predictable, and tailored for the specific food production situation. d Situations 1, 2, 3, and 4 for system output correspond to: no information (1) → absent, not applied, unknown; poor output (2) → ad-hoc sampling, minimal criteria used for evaluation, various food safety problems due to different problems in the activities; moderate output (3) → regular sampling, several criteria used for evaluation, restricted food safety problems mainly due to one (restricted) type of problem in the activities; good output (4) → systematic evaluation, using specific criteria, no safety problems.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Specificity sanitation program

Specificity maintenance program

Sophistication supplier control

Sophistication material control

Availability procedures

Hygienic performance

Sophistication translation requirements

Extent feedback info use

Sophisticaion validation of preventive measures

Sophisticaion validation of monitoring system

Sophisticaion validation of intervention measures

Extent verification people

Extent verification equipment

Appropriateness documentation

Appropriateness record keeping

Presence technical staff

Sufficiency workers' competences

Degree workers' involvement

Level of formalization

Sufficiency information system

Power in supplier relationships

Degree information exchange

Specificity external support

Specificity legal framework

Adequacy packaging equipment

Adequacy partial intervention

Packaging capability

Capability partial intervention

Adequacy analytical equipment pesticides

Sophisticaion validation of intervention measures

Severity stakeholder requirements

Sufficiency authorities

Extent personal hygiene requirements

Adequacy storage facilities

Adequacy irrigation method

Appropriateness hazard analysis

Storage capacity

Measuring equipment performance

Packaging capacity

PC1 (36.9%)

PC2 (7.3%)

PC3 (5.1%)

Figure 3: Loading plots of the first three principle components explaining 49.3% of variance

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Figure 4: PCA score plots for comparison between the clusters

-3

-2

-1

0

1

2

3

-3 -2 -1 0 1 2 3

PC

2 (

7.3

%)

Cluster 1

Cluster 2

Cluster 3

-3

-2

-1

0

1

2

3

-3 -2 -1 0 1 2 3

PC

3 (

5.1

%)

PC1 (36.9%)

PC1 (36.9%)

A

B

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPTHighlights:

• Assessment of safety management systems at primary production of fresh produce • 118 companies located in the European Union, emerging and developing countries

• Status of FSMS is affected by few main factors independent of location • International export supply chains promote capacity building

• Standards trigger maturation of FSMS